Publication

Abstract : The difficulty of recognizing emotions from pictures is a real problem for mental health evaluation and human-computer interaction. In this study, it is suggested that the use of Convolutional Neural Networks (CNNs) can help identify emotions in photographs focusing on classifying happy and sad states. The CNN architecture was trained using a dataset made up of images depicting people in various emotional states ranging from joy to sadness. After extensive testing and analysis, our model discerns between happiness and sadness. So, we investigate how the emotion recognition system performance could be maximized by trying out different CNN designs, data augmentation techniques as well as hyperparameters. Moreover, part of this chapter looks at the interpretability of the model’s decisions and also discusses implications and further research directions into emotion recognition with deep learning mechanisms. On the whole, our work advances emotion recognition technology and highlights how CNN-based techniques can aid in understanding human emotions through visual cues.